基于FTA与改进神经网络的轨道电路红光带诊断方法

以ZPW-2000无绝缘移频轨道电路红光带故障为研究对象,针对其故障的多样性与复杂性,提出一种基于故障树分析(FTA)与改进BP神经网络相结合的轨道电路智能故障诊断方法。根据轨道电路组成与故障关系建立故障树进行FTA定性分析,分析故障成因并提取故障诊断规则,确定诊断模型的输入输出,构建两个BP神经子网以并联方式联接组成诊断模型,采用LM算法和遗传算法对模型参数进行调整。通过仿真分析,表明该方法可行有效,为轨道电路智能故障诊断提供一种新思路。

Diagnosis of Red Tape of Track Circuit Based on FTA and Improved Neural Network

In view of the red tape faults of ZPW-2000 track circuit and the diversity and complexity of the faults,this paper proposes an intelligent track circuit fault diagnostic method based on Fault Tree Analysis( FTA) combined with improved BP neural network. The fault tree is established according to the track circuit components and fault relations to perform qualitative analysis of FTA,analyze fault causes,extract fault diagnosis rules and determine the input and output of the diagnosis model. Two parallel BP neural subnets are established and connected in parallel to compose diagnostic model,and model parameters are adjusted by LM and Genetic Algorithm. The simulation analysis proves that the method is feasible and effective,forging a new idea for intelligent fault diagnosis of the track circuit.